Factor structure and sex invariance of the temporal experience of pleasure scale (TEPS) in Chinese university students and clinical population

Anhedonia, defined as a diminished or absent ability to experience pleasure [1], is a core symptom of several psychiatric disorders, most prominently including major depressive disorder (MDD) and schizophrenia [2‐4]. In recent years, anhedonia has been linked to the increased risk of many other kinds of neuropsychiatric diseases, for example, Vaquero-Puyuelo D et al. found that anhedonia is a risk factor for Alzheimer’s disease in a longitudinal study [5]. Several scales aimed at evaluating anhedonia have been developed. Some initial measurements, such as the Fawcett-Clark Pleasure Capacity Scale [6] and the Snaith-Hamilton Pleasure Scale [7], developed in the early 1980s and 1990s, respectively, treat anhedonia as a unitary construct. Meanwhile, research providing deeper insights into anhedonia has indicated that anhedonia should be considered as a complex, multidimensional concept with relations to multiple psychopathological processes, including processes in physical/social, consummatory/anticipatory, and motivation/experiential dimensions [8, 9]. Scales with a focus on the physical/social dimension, including the Revised Physical Anhedonia Scale and the Revised Social Anhedonia Scale, have been used to assess anhedonia in patients with schizophrenic spectrum disorders and in other clinical samples since the 1970s [10, 11].

In the last two decades, there has been a growing acceptance of the concepts of anticipatory and consummatory anhedonia based on a convergence of evidence from animal and human studies [12‐14]. In response to this evolution in the field, Gard and colleagues developed the Temporal Experience of Pleasure Scale (TEPS), which was designed to reflect anticipatory and consummatory anhedonia [15]. In response to a recent growing emphasis on motivational aspects of anhedonia existing beyond the consummatory/anticipatory dimension [16, 17], the Motivation and Pleasure Scale (self-report) was developed to assess motivational anhedonia [18]. Although several scales assess anhedonia from a particular dimension perspective, a scale that encompasses the multi-dimensionality of anhedonia comprehensively has been lacking.

The TEPS has distinct advantages with respect to assessing anticipatory and consummatory anhedonia, and has been supported by substantial preclinical and nonclinical studies [19, 20]. It has also been translated into multiple language versions and demonstrated to be a reliable and valid scale for the measurement of anhedonia [21‐24]. However, there is inconsistency among studies with respect to the factor structure of the TEPS. The initial TEPS validation study reported by Gard and colleagues, which was conducted with a sample of American university students, yielded a two-factor structure with anticipatory and consummatory factors [15]. Subsequent studies of the French and German versions, conducted in both healthy controls and schizophrenic patients, replicated the two-factor structure proposed in the initial study [21‐23]. However, the Italian and Persian versions were found to have three-factor structures wherein it appeared to be more suitable to subdivide anticipatory anhedonia into contextual and abstract anticipatory anhedonia [25, 26]. In 2012, Chan et al. developed a Chinese version of the TEPS that included two additional items, altering the scale from an 18-item to a 20-item version, and then reported the psychometric properties of the new 20-item version in a non-clinical sample of undergraduate students [24]. Based on their findings, they proposed a four-factor structure for their 20- item TEPS, with the factors being consummatory contextual, consummatory abstract, anticipatory contextual, and anticipatory abstract [27]. The addition of two items, however, made it difficult to compare the properties found for the 20-item version with other international studies of 18-item versions of the TEPS. Moreover, it seems that the abstract-contextual dimension divisions were made primarily to suit to the anticipatory pleasure construct, with less relevance to the consummatory pleasure construct. For example, the “abstract” consummatory factor items may also refer to specific aspects of contextual pleasure (e.g. item 2: The sound of crackling wood in the fireplace is very relaxing). Therefore, the factor structure and interpretation of the Chinese TEPS, particularly of the original 18-item version, needs further exploration in non-clinical and clinical samples.

Despite debates regarding its factor structure, the TEPS has continued to be used worldwide in patients with psychiatric disorders, and has been reported to have good reliability and validity for assessing degrees of consummatory and anticipatory anhedonia [21, 23, 28]. For instance, in a schizophrenia sample, an 18-item German version of the TEPS was found to have a Cronbach’s α of 0.85 for the full scale, and scores correlated with scores from self-rated anhedonia and clinician-rated apathy measures [23]. Interestingly, impairments in consummatory pleasure and anticipatory pleasure appear to be distinguishable and distinct across different psychiatric disorders. Schizophrenics have been found to have reduced anticipatory pleasure, compared to healthy controls, but an intact capacity for consummatory pleasure [20, 29, 30], whereas patients with obsessive-compulsive disorder have reduced consummatory pleasure with intact anticipatory pleasure [31], and patients with MDD have reduced anticipatory and consummatory pleasure [32].

The aforementioned studies indicate that anhedonia is a multidimensional construction and that understanding the pathology of anhedonia involves distinguishing among its components. However, the question of how these differences may contribute to authentic differences between clinical and non-clinical samples remains to be clarified. Furthermore, there remains a need for analyses of the measurement invariance of the TEPS across different populations. Measurement invariance, which is an index of the stability of the meanings of scale items and factor structure across different populations, can be determined through a series of confirmatory factor analyses (CFAs) with increasingly constrained conditions enabling researchers to confirm that differences observed between groups reflect true differences between groups rather than an artifact of the scale being used [33].

Sex differences in anhedonia have also been reported between different psychiatric disorder populations. For example, schizophrenic male score higher than schizophrenic female on physical anhedonia and social anhedonia measures [34]. Additionally, trait anhedonia in female, but not male, alcoholics can predict depressiveness at the end of detoxification [35]. Although measurement invariance across sex has been reported for the TEPS in a healthy population sample [27], measurement invariance of the TEPS across sex in populations with psychiatric disorders has not been established. Such measurement invariance is important for the development of individualized clinical treatment plans in light of differing anhedonia scale findings between males and females.

The present study pursued four aims. Firstly, the reliability including internal consistency reliability and retest reliability of 18-item version of TEPS were explored in this study. Secondly, we used combined exploratory factor analysis (EFA) and CFA to determine the best-fit factor structure of a Chinese version of the 18-item TEPS in non-clinical and clinical samples. Thirdly, we explored the psychological mechanism of anhedonia revealed by that factor structure. Additionally, we used the best-fit factor structure to probe measurement invariance across clinical and non-clinical samples and across sex groups.

In the current study, a newly established four-factor structure was affirmed as the most suitable factor structure for a Chinese version of the 18-item TEPS in both clinical and non-clinical samples. In this four-factor structure, anticipatory pleasure and consummatory pleasure factors were each subdivided based on a motivation dimension. Measurement invariance analysis across clinical and non-clinical samples achieved strict invariance. Measurement invariance analysis across sex obtained strict invariance in the clinical sample and partial scalar invariance in the non-clinical sample.

In line with previous research, the TEPS was demonstrated to be a reliable, stable tool for assessing anhedonia in this study, with a Cronbach’α and McDonald’s omega > 0.80 and a retest coefficient of 0.70 [15, 24]. Furthermore, this study confirmed that the 18-item TEPS can be used to assess consummatory pleasure and anticipatory pleasure separately in Chinese respondents [15, 22]. Although item 11 and item 12 were transferred from anticipatory pleasure to consummatory pleasure in this study, the contents of these items contain a definite time and location, probing the ability of people to feel pleasure before anticipatory events have occurred. For example, item 11 (When I’m on my way to an amusement park, I can hardly wait to ride the roller coasters) suggests that being route to an amusement park can be enough to make someone feel pleasure. Consistent with previous research, we found that item 7 had low factor loading and thus decided to omit it in our CFA [19, 24]. A prior TEPS study in China also reported that item 7 had weak relations to others items and argued that there may be a cultural divergence in the interpretation of item 7 [24].

The consummatory-anticipatory division of anhedonia has gained substantial empirical evidence in animal studies as well as in human neurophysiological and behavioral studies [12, 56, 57]. There appears to be dissociable neural substrates mediating consummatory pleasure and anticipatory pleasure, with the former having been related to opioid neurotransmission onto GABAergic spiny neurons in the nucleus accumbens and the latter being related to mesolimbic dopaminergic neurotransmission [12]. Parsing anhedonia into finer components will benefit our understanding of the role of specific neurotransmitters and neural systems in mediating the psychological phenomenon of anhedonia. In addition, the full measurement invariance affirmed here across sex groups as well as across clinical versus non-clinical samples supports the view that the TEPS can be considered a stable assessment tool for the evaluation of consummatory pleasure and anticipatory pleasure. Furthermore, our comparison of goodness-of-fit indices among the initial two-factor structure of the TEPS [15], a previously published four-factor structure for a Chinese version of the TEPS with two added items [24], and the presently introduced four-factor structure that was newly developed through CFA indicates that our newly developed four-factor structure provides an improved factor structure in both non-clinical and clinical samples.

Regarding our division between with and without motivation-driven dimensions of consummatory and anticipatory pleasure, the role of motivation in anhedonia may involve value and effort computation for goal-directed behaviors, which are key in reward processing [17, 58]. Regarding motivation deficits related to anhedonia, someone who does not enjoy a typically rewarding activity due to anhedonia would be expected to lack motivation to pursue that activity in the future [59]. In turn, an individual with motivation deficits may experience less pleasure, favoring a vicious cycle between a motivation deficit and anhedonia. For example, compared to healthy controls, people with MDD tend to choose lower effort-expenditure rewarding tasks and expend less effort in a progressive ratio task [60]. Similarly, people with schizophrenia have been found to also tend to favor lower effort-expenditure rewarding tasks than controls, and their anhedonia symptoms have been shown to correlate with motivation deficit magnitude [61, 62]. Hence, it is important to have a scale that can be used to assess motivation in the context of anhedonia, and the presently developed four-factor structure of the TEPS suggests that the TEPS has the potential to be used in such an application.

The first factor identified, namely anticipatory pleasure with motivation-driving, reflects the need for approach motivation to achieve anticipated events, including making plans and anticipating effort (e.g., item 14, When I think of something tasty, like a chocolate chip cookie, I have to have one). Conversely, the second factor, anticipatory pleasure without motivation-driving, encompasses anticipatory events but with a limited need for motivation or an affirmative attitude (e.g., item 15, Looking forward to a pleasurable experience is in itself pleasurable). Meanwhile, the third factor, consummatory pleasure with motivation-driving, refers to pleasurable experiences in which one must be motivated to actively remember a pleasure to activate and maintain a representation (e.g., item 9, I love it when people play with my hair). Lastly, the fourth factor, consummatory pleasure without motivation-driving, refers to immediately accessible enjoyment without a need to desire more. Constituent items of this fourth factor tend to refer to natural events involving limited subjective initiative (e.g., item 13, I appreciate the beauty of a fresh snowfall).

The motivation dimension affirmed in our analyses was found to exist for both consummatory pleasure and anticipatory pleasure, a supposition that is supported by previous empirical evidence. For example, consummatory and anticipatory pleasure were found to be related to motivation-associatedevent-related potential signals in a study employing a cue gambling task [63]. Meanwhile, significantly impaired consummatory and anticipatory pleasure in schizophrenia may only emerge in the context of a serious motivation deficit [64]. Hence, the consideration of motivation in the factor structure of the TEPS may provide new insight enabling a more comprehensive understanding of the TEPS, thus expanding the applicability of the TEPS in relation to a more detailed structural understanding of anhedonia.

The newly explored four-factor structure showed good convergent validity and discriminate validity. Previous studies mainly focused on the initial two-factor structure’s convergent and discriminate validity through correlation analysis. In the initial study, Grad et al. demonstrated that the two-factor structure have good convergent validity and discriminate validity. The result of correlation analysis indicated that both anticipatory pleasure factor and consummatory pleasure factor significantly correlated with other scales measuring anhedonia and showed separation correlation patterns as well [15]. Subsequently, Paul et al. not only demonstrated that the two-factor structure have good convergent validity and discriminate validity, but also indicated that TEPS can possibly assess motivation components of anhedonia [19]. Although prior studies only demonstrated that the two-factor structure showed good convergent validity and discriminate validity, this study also showed that the newly explored four-factor structure has good convergent validity and discriminate validity through CR and HTMT indexes.

The present findings of full measurement invariance (configural, metric, scalar, and error variance) for our newly explored four-factor structure across clinical and non-clinical samples indicate that differences in TEPS performance between the clinical and non-clinical respondents can be interpreted reliably. This affirmation is important given that TEPS scores have been reported to differ between clinical and non-clinical groups and appear to be differentially affected across clinical groups (e.g. impaired anticipatory pleasure and consummatory pleasure in MDD [32], but bias toward impairment in anticipatory pleasure in schizophrenia [20, 29‐31]). Furthermore, recent studies employing reward task-choice behavioral analyses have suggested that there may be distinct clinically important motivation impairments in relation to anhedonia in patients with schizophrenia, bipolar disorder, and MDD [65, 66]. Hence, measurement invariance provides confidence that TEPS score differences reflect authentic differences across the well-establishedanticipatory-consummatory pleasure dimension as well as across a newly exposed motivation-nonmotivation dimension in future research.

With respect to measurement invariance across sex in a non-clinical sample, based on modification indices, we permitted correlations of the error variance between item 2 (The sound of crackling wood in the fireplace is very relaxing) and item 4 (I love the sound of rain on the windows when I’m lying in my warm bed). These two items both contribute to the consummatory pleasure without motivation-driving subscale and reflect pleasure related to a natural phenomenon, which may be related to a residual correlation between them. The intercepts of items 2, 9, 11, and 18 were free estimated when we conducted scalar invariance analysis, indicating that researchers should further test whether these items perform differently in relation to sex in a non-clinical sample. Similarly, Zhou et al. also reported partial scalar invariance across sex in an undergraduate sample employing a 20-item version of the TEPS [27]. Notwithstanding, the full measurement invariance observed across sex in our clinical sample may be more important in the context of exploring the mechanisms of anhedonia. With respect to sex differences in anhedonia [34, 35], some researchers have reported that men are more likely to experience anhedonia than women [67]. Indeed, a meta-analysis indicated that, compared to females, males with schizophrenia spectrum disorders report lower anticipatory pleasure and consummatory pleasure, as measured by the TEPS [30]. Overall, affirmation of measurement invariance provides a premise for probing differences in anhedonia pattern between males and females.

Some limitations of this study should be noted. First, because this study was conducted in a Chinese cultural context, the generalizability of the findings to other cultural contexts remains to be determined. Using a 20-item version of the TEPS in China, Chan et al. did find full measurement invariance for consummatory pleasure and anticipatory pleasure subscales across cultures, which may offset this limitation to some [68]. Second, our enrollment of college students as a non-clinical sample may restrict the generalizability of our findings to other age ranges. With respect to age generalization, however, prior studies examining the TEPS in community samples have affirmed its stability across age bands [21, 23]. Third, the cut-off points of indices used for CFA and TLI may be too lax for the assessment of the goodness of fit in this study, as recently some researchers proposed considered values over .95 for CFI and TLI and below .08 for SRMR and .06 for RMSEA [49], or values above .95 or .97 for CFI and TLI paired with values lower than .05 for SRMR and RMSEA as representative of a good fit with data [69]. Even Sivo et al. considered sample size to establish cut-off points [70], which means that depending on the sample size of this study, values higher than .99 for CFI and TLI accompanied by values lower than .05 and below .03 for RMSEA are indicative of a good fit to data. However, it also should be noticed that the fit indices of new four-factor found in this study achieved traditional criteria of the goodness of fit, and better than others two models. Lastly, as Fan and Sivo suggested, the Δgoodness-of-fit indexes are designed to assess model fit in terms of covariance structure and may be sensitive to model size [71], which should be considered especially when mean structure invariance is the research focus.

For the first time, this study determined the most suitable factor structure for a Chinese version of the 18-item TEPS with multiple methods and further confirmed the resultant four-factor structure in clinical and non-clinical samples. Besides, the newly four-factor structure was demonstrated with good convergent validity and discriminate validity. Full measurement invariance was observed between clinical and non-clinical samples and between sex in a clinical sample. Partial scalar invariance was found across sex in a non-clinical sample. The present results provide insights into the psychometric properties of the TEPS as well as the psychological phenomenon of anhedonia.